Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2023 Sep 25.
Published in final edited form as: Neurobiol Dis. 2023 Jun 10;184:106192. doi: 10.1016/j.nbd.2023.106192

Epigenomic Analysis of Alzheimer’s Disease Brains Reveals Diminished CTCF Binding on Genes Involved in Synaptic Organization

Prachetas J Patel 1, Yong Ren 1, Zhen Yan 1,*
PMCID: PMC10519202  NIHMSID: NIHMS1931430  PMID: 37302762

Abstract

Epigenetic aberrations are suggested to play an important role in transcriptional alterations in Alzheimer’s disease (AD). One of the key mechanisms of epigenetic regulation of gene expression is through the dynamic organization of chromatin structure via the master genome architecture protein, CCCTC-binding factor (CTCF). By forming chromatin loops, CTCF can influence gene transcription in a complex manner. To find out whether genome-wide DNA binding sites for CTCF are altered in AD, we compared CTCF chromatin immunoprecipitation sequencing (ChIP-Seq) data from frontal cortex of human AD patients and normal controls (n=9 pairs, all females). We have revealed that CTCF-binding affinity on many genes is significantly reduced in AD patients, and these genes are enriched in synaptic organization, cell adhesion, and actin cytoskeleton, including synaptic scaffolding molecules and receptors, such as SHANK2, HOMER1, NRXN1, CNTNAP2 and GRIN2A, and protocadherin (PCDH) and cadherin (CDH) family members. By comparing transcriptomic data from AD patients, we have discovered that many of the synaptic and adhesion genes with reduced CTCF binding in AD are significantly reduced in their mRNA expression. Moreover, a significant overlap of genes with the diminished CTCF binding and the reduced H3K27ac is identified in AD, with the common genes enriched in synaptic organization. These data suggest that the CTCF-controlled 3D chromatin organization is perturbed in AD, which may be linked to the diminished expression of target genes, probably through changes in histone modification.

Keywords: Alzheimer’s disease, epigenetics, CTCF, chromatin organization, synaptic genes, histone acetylation

1. Introduction

Alzheimer’s disease (AD) is a complex neurodegenerative disorder with multiple pathological features, including synapse loss, synaptic and neuronal network dysfunction (Selkoe, 2002; Tracy and Gan, 2018; Wilson et al., 2023). Large-scale transcriptomic studies have revealed numerous genes with altered expression in postmortem prefrontal cortex (PFC) of AD patients (Mathys et al., 2019; Zhang et al., 2013). However, it is unclear how the aberrations in gene transcription occur in AD.

Epigenetic control, which largely involves chromatin remodeling, is a key avenue to achieve transcriptional regulation of gene expression in neurons (Borrelli et al., 2008). Epigenetic alterations are found to be a primary hallmark and causal driver of aging (López-Otín et al., 2023; Yang et al., 2023), the most prominent contributing factor of AD. Mammalian genomes are packaged into hierarchical levels, including active and inactive chromatin compartments at the mega-base scale and topologically associating domains (TADs) at the sub-megabase scale (Dixon et al., 2012; Lieberman-Aiden et al., 2009). Genomic function is significantly influenced by chromatin organization in the three-dimensional nuclear space (Fraser and Bickmore, 2007; Lanctôt et al., 2007). The multifunctional zinc finger protein, CCCTC-binding factor (CTCF), is identified as a master organizer of chromatin architecture, which plays an essential role in mediating intra- and inter-chromosomal contacts (Merkenschlager and Duncan, 2013; Ong and Corces, 2014; Phillips and Corces, 2009). CTCF binds to a wide range of variant sequences to form genome-wide chromatin loops (Nora et al., 2017; Rao et al., 2014), coordinating long-range interactions between regulatory elements to control gene expression (Dehingia et al., 2022; Kubo et al., 2021; Merkenschlager and Odom, 2013; Ohlsson et al., 2010; Phillips and Corces, 2009).

De novo mutations in CTCF have been associated with various diseases, including Intellectual Disability, Schizophrenia, and cancer (Dehingia et al., 2022; Konrad et al., 2019; Ohlsson et al., 2001). Previous in vitro studies found that CTCF bound with a high affinity at the promoter region of Amyloid β-Protein Precursor, serving as a transcriptional activator (Vostrov and Quitschke, 1997). Here we examined whether CTCF occupancy was altered in AD brains. By bioinformatic analyses of epigenomic data from postmortem brain tissues of AD patients, we uncovered a significant alteration of CTCF binding on a genome-wide scale. Moreover, we found that the genes with reduced CTCF occupancy in AD were enriched in cell adhesion and synaptic organization, and their transcriptional expression was also significantly diminished in AD.

How could the diminished CTCF binding be associated with the reduced transcription of synaptic genes in AD? CTCF occupancy and CTCF-mediated chromatin loops can promote transcription initiation by recruiting RNA polymerase II (Chernukhin et al., 2007), preventing DNA methylation (Engel et al., 2006), marking boundaries of histone methylation domains (Barski et al., 2007), or regaining histone acetylation in postmitotic cells (Kang et al., 2020). From bioinformatic analyses of the ChIP-seq data of histone H3 lysine 27 acetylation (H3K27ac, an active enhancer mark linked to transcriptional activation), we found that genes with reduced H3K27ac in AD overlapped significantly with those with the diminished CTCF binding, and the common genes were enriched in synaptic organization. These data suggest that CTCF-mediated higher-order chromatin structure is altered in AD, which may contribute to the diminished histone acetylation and transcriptional activation of synaptic genes.

2. Material and Methods

2.1. ChIP-seq Data Samples

CTCF and H3K27ac ChIP-seq data were acquired from ENCODE consortium’s Rush Alzheimer’s Disease study (ENCODE Project Consortium, 2012; Luo et al., 2020). Single-end ChIP-seq performed by the Bernstein Lab on human postmortem brain tissue (middle frontal area 46) with AD and control subjects (no Cognitive Impairment) was used for this study. We analyzed female AD (n=9) and female control (n=9) samples at the time of data availability. Both CTCF ChIP-seq and H3K27ac ChIP-seq data were derived from the same human samples. All sample information (ENCODE accession numbers, age, sex etc.) is available in Supplementary Table 8.

2.2. ChIP-seq Differential Peak Analysis

Differential analysis was performed using DiffBind R package (v 3.2.5) (Ross-Innes et al., 2012). EdgeR methodology was used for comparing ChIPseq data that were normalized with ‘trimmed mean of M values’ (TMM) (Chen et al., 2016; McCarthy et al., 2012; Robinson et al., 2010). The cutoff threshold for significantly different peaks across groups was set at p-value ≤ 0.05 (two-sided Wilcoxon ‘Mann-Whitney’ test). ChIPseeker R package (Wang et al., 2022; Yu et al., 2015) annotated the identified ChIP-seq peaks with associated gene names derived from relative location on the human genome (hg38, promoter defined as ± 3kb of transcription start site). Comparison heatmap of CTCF peak profiles of AD and Control groups and distribution of differential CTCF peaks were also generated using ChIPseeker. BigWig files of ChIP-seq data were generated with DeepTools (Ramírez et al., 2014) Python package and BedTools (Quinlan and Hall, 2010) UNIX suite. ChIP-seq landscapes were generated using BigWig files and Integrative Genomics Viewer (IGV) Web App [https://igv.org] (Robinson et al., 2011). H3K4me3 peaks were used as markers for gene promoter locations (Sharifi-Zarchi et al., 2017) and H3K27ac peaks were used to mark gene enhancer locations (Creyghton et al., 2010). The 3D Genome Browser and the available CTCF Hi-C contact map (Wang et al., 2018) were used to locate topologically associated domains (TADs).

2.3. Gene Ontology (GO) Analyses, Protein-protein Interactions (PPI) Network and Hub Genes

Metascape [https://metascape.org] (Zhou et al., 2019) was used to identify enriched categories of the genes with differential CTCF peaks in AD using default threshold parameters (overlap ≥ 3, p-value ≤ 0.01, enrichment ≥ 1.5). Metascape’s Gene Prioritization by Evidence Counting (GPEC)(beta) machine learning algorithm was used to rank the enriched pathways. SynGO [https://www.syngoportal.org/] (Koopmans et al., 2019) analysis was used to identify the enrichment (p-value ≤ 0.01) of synaptic genes from top GO pathways. String database [https://string.db] (Szklarczyk et al., 2021) was used to create PPI network for genes in top enriched categories. CytoHubba (Chin et al., 2014) app in Cytoscape (version 3.9.0) (Shannon et al., 2003) was used to detect top ranking hub genes using the Maximal Clique Centrality algorithm.

2.4. Overlapping Differential ChIP and Gene Expression

Microarray data from the dorsolateral PFC of 129 AD patients and 101 non-demented healthy controls (Harvard Brain Tissue Resource Center) (Zhang et al., 2013) were used for transcriptomic analysis. We used box plots generated by the ‘AD Consensus Trancriptomics’ ShinyApp (developed at Swarup lab by Sam Morabito) to compare microarray gene expression levels in control and AD samples (Morabito et al., 2020). InteractiveVenn [http://www.interactivenn.net/] (Heberle et al., 2015) was used to detect overlapping genes among gene sets acquired from differential ChIP-seq analysis and differential gene expression analysis. Heatmap of raw gene expression from the microarray data was generated using Phantasus [https://ctlab.itmo.ru/phantasus].

2.5. Chromatin Immunoprecipitation-PCR (ChIP-PCR)

Postmortem human frontal cortex (Brodmann’s area 10) from patients with AD (both males and females, Braak Stage V or VI) and age-matched control subjects were provided by the National Institutes of Health (NIH) NeuroBioBank. Upon arrival, tissue was stored in a −80°C freezer until use.

Human tissues were homogenized in 250 μl ice-cold douncing buffer (10 mM Tris-HCl, 4 mM MgCl2, 1 mM CaCl2, pH 7.5). Chromatin shearing used two approaches. In some experiments, the homogenized sample was incubated with 12.5 μl micrococcal nuclease (final conc. 5 U/ml, Sigma, N5386) for 7 min and terminated by adding EDTA (final conc. 10 mM). In other experiments, the homogenized sample was sonicated with Sonic Dismembrator Model 500 (Fisher Scientific) on ice (30% strength, 30-sec each time, 5 times). Both methods gave similar results, so data were pooled together. After chromatin shearing, hypotonic lysis buffer (1 ml) was added and incubated on ice for 1 hr. The supernatant was transferred to a new tube after centrifugation. After adding 10× incubation buffer (50 mM EDTA, 200 mM Tris-HCl, 500 mM NaCl), 10% of the supernatant was saved to serve as input control. To reduce nonspecific background, the supernatant was pre-cleared with 100 μl of salmon sperm DNA/protein A agarose-50% slurry (Millipore, 16-157) for 2 hrs at 4°C with agitation. The pre-cleared supernatant was incubated with antibodies against CTCF (3 μg per reaction; 07-729, Millipore/Sigma) overnight at 4°C under constant rotation, followed by incubation with 60 μl of Salmon Sperm DNA/Protein A agarose-50% Slurry for 2 hrs at 4°C. After washing for five times, bound complex was eluted twice from the beads by incubating with the elution buffer (100 μl) at room temperature. Proteins and RNA were removed by using proteinase K (Invitrogen) and RNase (Roche). Immunoprecipitated DNA and input DNA were purified by QIAquick PCR purification Kit (Qiagen). Quantification of ChIP signals was calculated as percent input. Purified DNA was subjected to qPCR reactions with primers against human GRIN2A enhancer (Forward, 5’ CCACAGGTCCTACTCTTCCC 3’; Reverse, 5’ GTGGTCATGGTTTCGCTGTC 3’), GRIN2A promoter (Forward, 5’ GAGCCCTGATCTCCCTCTG 3’; Reverse, 5’ CTAAACCCGGTGGCTGCT 3’), SHANK2 enhancer (Forward, 5’ CTCCAGGGGCTTTGTTTGTC 3’; Reverse, 5’ TGCCCACTTTGCTAGAGTCA 3’), and SHANK2 promoter (Forward, 5’ CCTGGATGGGAGCTCTGC 3’; Reverse, 5’ GTCCTGCCCCTTCCTCTGGAA 3’).

2.6. Statistical Analyses

ChIP-seq Differential Analysis: statistically significant peaks in Diffbind (measured by difference in read densities) were identified by EdgeR on a negative binomial distribution model to estimate variance. All peaks annotated were ranked by p-values that were computed using a two-sided Wilcoxon ‘Mann-Whitney’ test. CTCF peaks at individual genes across sample groups were compared using unpaired t-test with Welch’s correction.

Gene Ontology Pathway Enrichment Analysis: Metascape’s gene set enrichment analysis uses the following equation to calculate the enriched pathways’ p-values derived from hypergeometric distribution. This p value is further corrected by Benjamini Hochberg correction to get False Discovery Rate (FDR).

p=∑i=nmin(M,K)(in)(N−kM−i)

Where N is total genes in human genome, M is total input genes, k is genes in selected pathway for analysis, and n is the genes common in k and M.

SynGO enrichment analysis was based on a combination of “Generalized Gene-Set Analysis of GWAS Datagene-set analysis (MAGMA)” (De Leeuw et al., 2015), “Stratified LD-Score Regression (S-LDSC)” (Finucane et al., 2015; Gazal et al., 2017) using brain expressed GWAS datasets (Koopmans et al., 2019).

Protein-Protein Interaction Network Analysis: STRING PPI analysis was used with medium confidence set as the cut-off for selected gene sets. Genes with no edges were removed. Modules were defined using gene enrichment analysis. To determine hub genes, we used CytoHubba’s MCC setting, which is based on how essential proteins are clustered. MCC of a node v=∑C∈S(v)(|C|−1)!, where S(v) = collection of maximal cliques containing v, and (|C|−1)! contains all positive integers less than |C|.

Microarray Differential Expression Analysis: Differential expression for microarray data was tested for statistical significance by limma classic t-statistic, and arranged by fold change, p-value and adjusted p-value (Benjamini-Horchberg).

Overlapping of Gene Lists: Hypergeometric test was used to calculate the p-value for under- or over-enrichment based on the cumulative distribution function (CDF). Expected number of successes = (s*M)/N, s: sample size, M: number of successes in the population, N: population size. If k (actual number of successes) > (s*M)/N, then k is considered over-enriched, otherwise under-enriched.

3. Results

3.1. AD Patients Have Reduced CTCF Binding Affinity Genome-wide in Postmortem Brains.

We performed epigenomic analysis of CTCF ChIP-seq data of postmortem brain tissues (BA46 of prefrontal cortex) in control and AD patients (n=9 pairs, all females) acquired from the ongoing ENCODE project by Rush University Alzheimer’s Disease Center (ENCODE Project Consortium, 2012). Using DiffBind R Bioconductor package, we found that 1742 out of 53,377 CTCF-binding peaks were significantly altered in AD samples (Supplementary Table 1), with 1719 reduced CTCF peaks and 23 increased CTCF peaks. While the majority of differential CTCF peaks showed the loss in AD samples, compared to controls (Figure 1A), H3K27ac peaks showed both gain and loss in AD (Figure 1B), suggesting that the marked reduction of CTCF occupancy in AD is not due to tissue qualities. The significant reduction of CTCF binding in AD was found at gene bodies, gene promoters and intergenic regions, and 24.4% was at the promoter (3kb around TSS) (Figure 1C). This reduction of CTCF binding on target genes in AD (Figure 1D) could lead to changes in CTCF-mediated chromatin looping and genome architecture, resulting in altered gene expression.

Figure 1, Genome-wide CTCF binding is significantly diminished in frontal cortex of AD patients.

Figure 1,

Open in a new tab

A, B, MA plot depicting fold changes over binding affinities for CTCF (A) or H3K27ac (B) in AD patients (n=9), compared to control samples (n=9). Pink dots indicate significantly differentially bound sites (p ≤ 0.05, two-sided Wilcoxon ‘Mann-Whitney’ test). C, Pie chart illustrating the locations of differentially bound CTCF sites in AD patients on the genome. D, CTCF ChIP-seq profile plots and heatmaps illustrating the significantly reduced CTCF binding sites (± 3Kb from peak centers) in AD patients, compared to control samples.

3.2. Genes with Reduced CTCF Binding in AD Are Enriched in Cell Junction Organization.

Gene Ontology (GO) enrichment analyses of the 1241 genes with significantly reduced CTCF binding in AD indicated that the most prominently enriched biological processes are involved in the regulation of neuron projection, actin filaments, receptor signaling, cell adhesion and synaptic transmission (Figure 2A, Supplementary Table 2), and the most enriched cellular components are postsynapse, axon, presynapse, actin cytoskeleton, and cell-cell junction (Figure 2B).

Figure 2, Genes with the significantly reduced CTCF binding in AD are enriched in synaptic organization, cell adhesion and actin cytoskeleton.

Figure 2,

Open in a new tab

A, B, Gene Ontology (GO) analysis with Metascape depicting the most enriched categories in Biological Processes (A) or Cellular Component (B) of all the genes with significantly reduced CTCF binding in AD patients. C, Protein-Protein Interaction (PPI) networks of genes with significantly reduced CTCF binding in AD among the top 8 GO categories. HUB genes are highlighted with red. D, Sunburst plot from SynGO analysis illustrating the most enriched categories in Biological Processes (top) or Cellular Component (bottom) of the synaptic genes with significantly reduced CTCF binding in AD among the top 8 GO categories. Higher red intensities are associated with more significant enrichments.

Protein-Protein Interaction (PPI) network analyses indicated that genes with significantly reduced CTCF binding in AD are highly clustered in cell junction organization and actin cytoskeleton (Figure 2C). Hub gene analysis was performed to identify key genes that represent the central molecular constituents most strongly connected in the gene network, which are presumably most responsible for the network’s global function. Using CytoHubba, which ranks proteins based on node features in the network, we found that hub genes with the highest rankings were associated with synaptic organization, including SHANK1/2, HOMER1 and DLG2 (encoding scaffolding/anchoring proteins at glutamatergic synapses), NRXN1 (encoding the pre- and post-synaptic membrane cell-adhesion molecule Neurexin 1), ERBB4 (encoding receptor tyrosine kinase erbB-4 activated by Neuregulins), GRIN2A (encoding NMDAR subunit NR2A), and GRM3 (encoding metabotropic glutamate receptor 3). Other notable synaptic genes with the reduction of CTCF binding include CNTNAP2 (encoding a Contactin-associated protein in the Neurexin family), NRXN3 (encoding Neurexin 3), ERBB3 (encoding erbB-3), STXBP6 (encoding Syntaxin Binding Protein 6), and KALRN (encoding Kalirin RhoGEF kinase that is important for spine formation) (Xie et al., 2007). One cluster of genes with reduced CTCF binding in AD, which is enriched in “cell junction organization & adhesion” category, was the protocadherin (PCDHB9, PCDHB13, PCDHB14) and cadherin (CDH8, CDH13, CDH19, CDH24) family members.

Among the genes with reduced CTCF binding in AD, which are enriched in the category of “cellular response to stress”, many are involved in chromosome organization and DNA repair, like the hub genes POLE and POLE4 (encoding a DNA polymerase), RFC2 (encoding a replication factor required by DNA polymerases for DNA elongation) and ERCC (encoding a DNA excision repair protein). In addition, many histone modifiers and chromatin remodelers were also found in this group, including HDAC9, HAT1, KAT6B, KMT2A/C, KDM4B, SUV420H2, SETD7, DOT1L, PHF8, SMARCB1, and ARID1A.

With cell junction organization being the most significantly enriched GO category for genes with reduced CTCF binding in AD, we utilized SynGo, a synaptic gene ontology database, to further explore dysregulated synaptic pathways in Alzheimer’s disease. Among the genes with reduced CTCF binding, 196 were classified as synaptic genes (Supplementary Table 3). Synaptic enrichment analyses identified synaptic organization, transport and signaling as the most enriched biological processes, while presynapse and postsynapse as the most enriched and abundant subcellular components, for synaptic genes with reduced CTCF binding in AD (Figure 2D).

3.3. Reduced CTCF Binding is Linked to Altered Gene Expression in AD.

To find out whether the reduction of CTCF binding on target genes in AD is linked to their transcriptional alteration, we compared ChIP-seq data with transcriptomic analysis of microarray data from the dorsolateral PFC of 129 AD patients and 101 non-demented healthy controls (Harvard Brain Tissue Resource Center) (Zhang et al., 2013). With a cutoff of adjusted p < 0.001 and fold change (FC) of 10%, we identified 1516 downregulated genes in AD patients, which are enriched in synaptic signaling, synapse organization, ion transport, neurotransmitter secretion, and glutamatergic synaptic transmission (Williams et al., 2021). Among genes with reduced CTCF binding (1241) and genes with downregulated expression (1516) in AD, 101 overlapped genes were identified in the two groups (Figure 3A, Supplementary Table 4). GO analyses of these common genes indicated that they were enriched in Biological Processes like axon guidance, regulation of neural transmission, trans-synaptic signaling, postsynaptic organization and synaptic adhesion, and Cellular Components like pre- and post-synaptic membrane and transmembrane transporter complex (Figure 3B). PPI of these common genes in top GO pathways pointed to key synaptic genes, such as GRIN2A, SHANK2, NRXN1, SNAP91, and CNTNAP2 (Figure 3C).

Figure 3, Genes with reduced CTCF binding in AD show altered mRNA expression in AD.

Figure 3,

Open in a new tab

A, D, Venn diagrams showing the overlap of genes with reduced CTCF binding in AD and genes with downregulated (A) or upregulated (D) expression in AD from Microarray data. B, E, Gene Ontology (GO) showing the enriched pathways of overlapped genes with reduced CTCF binding and down-regulated (B) or up-regulated (E) expression in AD. BP: Biological Processes; CC: Cellular Component. C, PPI networks of overlapped genes with reduced CTCF binding and down-regulated expression in AD. F, Heatmaps showing the expression of selected genes with reduced CTCF binding in AD in top GO pathways (synaptic organization, cell adhesion, and actin cytoskeleton) in control (n=101) vs. AD (n=129) samples. Red: Down-regulated genes; Black: non-changed genes; Green: Up-regulated genes.

On the other hand, we identified 1134 upregulated genes in AD patients, which are enriched in immune response activation and cell death (Williams et al., 2021). Only 64 overlapped genes were identified between genes with reduced CTCF binding (1241) and genes with upregulated expression (1134) in AD (Figure 3D). These common genes were enriched in T-cell activation, MAPK cascade and actin cytoskeleton polymerization (Figure 3E).

Next, we examined the expression of selected genes with reduced CTCF binding in the 3 major GO pathways: synaptic organization, cell adhesion, actin cytoskeleton. As shown in Figure 3F, compared to controls (n=101), most of the synaptic and adhesion genes exhibited a consistent reduction of the mRNA expression value in AD patients (n=129), including GRIN2A, NRXN1, HOMER1, SNAP91, KALRN, ERBB4, CNTNAP2, CDH8, CDH13, PCDHB12, PCDHB14, while actin cytoskeleton genes exhibited 3 patterns (DOWN, No Change, UP). These data suggest that the diminished CTCF binding on synaptic and adhesion genes may contribute to their transcriptional reduction in AD.

3.4. Genes in Cell Adhesion and Synaptic Organization Have Reduced CTCF Binding and Expression in AD.

To better understand the relationship between CTCF binding and gene expression, we further examined the ChIP-seq and transcriptomic data of a few key genes involved in cell adhesion or synaptic organization. As shown in representative landscapes of CTCF occupancies at the clustered protocadherin gene locus from a control subject and an AD patient (Figure 4A), CTCF peaks at PCDHα, PCDHβ and PCDHγ family members were markedly reduced in the AD sample. Significant differences of CTCF occupancy at PCDHα, PCDHβ and PCDHγ were observed between control and AD patients (n=9/group) (Figure 4B). Similarly, CTCF occupancies at CDH8, CDH13 and CDH19 were significantly reduced in AD patients, compared to controls (Figure 4C and 4D). Microarray data (Figure 4E) also revealed the significant reduction of protocadherin genes (PCDHα6, PCDHβ12) and cadherin genes (CDH8, CDH13) in AD, while CDH19 expression was not reduced in AD patients.

Figure 4, Cell adhesion genes in protocadherin and cadherin families show the reduced CTCF binding and expression in AD.

Figure 4,

Open in a new tab

A, C, Epigenome browser snapshots of ChIP-seq data showing the landscape of CTCF peaks at the clustered protocadherin gene locus encompassing PCDHα, PCDHβ and PCDHγ family members (A) or at cadherin 8 (CDH8) gene locus (C) in PFC from a control and an AD patient. The landscape of H3K4me3 and H3K27ac binding sites are also shown to illustrate the promoter and enhancer regions. B, D, Box plots showing CTCF peak scores from differential peak calling analysis with DiffBind at PCDHα, PCDHβ and PCDHγ (B) or cadherin genes including CDH8, CDH13, and CDH19 (D) in control vs. AD samples (n=9 pairs, unpaired t-test with Welch’s correction). E, Box plots showing mRNA expression levels of selected protocadherin and cadherin genes in control vs. AD samples from microarray data (control: n=101; AD: n=129, unpaired t test with Welch’s correction).

Hi-C probing of 3D genome architecture has identified ∼10,000 CTCF-bound loops that typically occur at contact domain boundaries, which frequently link promoters and enhancers, influencing gene activation (Rao et al., 2014). Hi-C interaction matrices at a topologically associating domain (TAD) encompassing GRIN2A locus show the intense contacts demarcated by CTCF-bound loops within the gene body and with other neighboring genes (Figure 5A). CTCF peaks at GRIN2A promoter (marked by H3K4me3), two enhancers (marked by H3K27ac), and a UTR region were significantly reduced in AD samples (Figure 5B and 5C). We then performed ChIP-PCR assays of postmortem PFC from AD patients to validate whether CTCF binding on GRIN2A was indeed decreased. As shown in Figure 5D, the CTCF peak score was significantly lower at the enhancer and promoter regions of GRIN2A in AD patients, compared to control subjects, consistent with the CTCF ChIP-seq epigenomic data (ENCODE Project Consortium, 2012). Concomitant with the reduction of CTCF binding on GRIN2A, the expression level of GRIN2A (Figure 5E) was significantly lower in AD samples (n=129), compared to controls (n=101).

Figure 5, Synaptic gene GRIN2A has the reduced CTCF binding and expression in AD.

Figure 5,

Open in a new tab

A, Heatmaps of contact matrices from Hi-C data at a topologically associating domain (TAD) encompassing GRIN2A locus. CTCF binding sites on GRIN2A are labeled with red pins. Each curved line connects two noncontiguous DNA segments. B, Epigenome browser snapshots of ChIP-seq data showing the landscape of CTCF peaks at GRIN2A gene locus in a control and an AD patient. The landscape of H3K4me3 and H3K27ac are also shown to illustrate the promoter and enhancer regions. Zoomed in CTCF peaks at specific loci are shown below. C, Box plots showing CTCF peak scores from differential peak calling analysis with DiffBind at the promoter and enhancer regions of GRIN2A in control vs. AD samples (n=9 pairs, unpaired t-test with Welch’s correction). All the p values are from unpaired t test with Welch’s correction. D, Bar graphs showing ChIP-PCR quantification of CTCF occupancy at GRIN2A enhancer 1 (E) or promoter (P) in PFC from control vs. AD patients (n=6/group, p=0.024 (E), p=0.042 (P), unpaired t-test with Welch’s correction). IgG controls are also shown to demonstrate the specificity of the ChIP-PCR data on CTCF binding. E, Box plots showing mRNA expression levels of GRIN2A in control vs. AD samples from microarray data (control: n=101; AD: n=129, unpaired t test with Welch’s correction).

Hi-C contact map also reveals the contacts at a TAD encompassing SHANK2 locus, which are composed of two separate clustered CTCF-bound chromatin loops (Figure 6A). A marked reduction of CTCF binding at the promoter and enhancer of SHANK2 was found in AD patients (Figure 6B and 6C). ChIP-PCR assays validated the significant decrease of CTCF peaks at SHANK2 enhancer and promoter in AD patients (Figure 6D). Moreover, the expression level of SHANK2 (Figure 6E) was significantly lower in AD samples. Taken together, these data suggest that the significantly diminished CTCF binding on adhesion and synaptic genes in AD may be responsible for their transcriptional repression due to the attenuation of chromatin looping to bring together promoters and enhancers.

Figure 6, Synaptic gene SHANK2 has the reduced CTCF binding and expression in AD.

Figure 6,

Open in a new tab

A, Heatmaps of contact matrices from Hi-C data at a TAD encompassing SHANK2 locus. CTCF binding sites on SHANK2 are labeled with red pins. B, Epigenome browser snapshots of ChIP-seq data showing the landscape of CTCF peaks at SHANK2 gene locus in a control and an AD patient. C, Box plots showing CTCF peak scores at the promoter and enhancer regions of SHANK2 in control vs. AD samples (n=9 pairs, unpaired t-test with Welch’s correction). D, Bar graphs showing ChIP-PCR quantification of CTCF occupancy at SHANK2 enhancer (E) or promoter (P) in PFC from control vs. AD patients (n=6/group, p=0.017 (E), p=0.001 (P), unpaired t-test with Welch’s correction). IgG controls are also shown. E, Box plots showing mRNA expression levels of SHANK2 in control vs. AD samples from microarray data (control: n=101; AD: n=129, unpaired t test with Welch’s correction).

3.5. Genes with Reduced CTCF Binding and Reduced Histone Acetylation in AD Converge on Synaptic Pathways.

Next, we further explored the potential molecular mechanism underlying the link of synaptic genes with reduced CTCF occupancy in AD to their diminished mRNA expression. CTCF is known to be able to control gene expression via diverse avenues, including the recruitment of transcriptional machinery and the regulation of DNA methylation or histone modifications (Chernukhin et al., 2007) (Barski et al., 2007; Engel et al., 2006; Kang et al., 2020). Thus, we examined ChIP-seq data of H3K27ac (an active enhancer mark) from postmortem PFC of control and AD patients (n=9 pairs) acquired from the ENCODE project (ENCODE Project Consortium, 2012), and compared them with CTCF ChIP-seq data. We found that 788 H3K27ac peaks were significantly reduced and 542 H3K27ac peaks were significantly increased in AD samples (fold change: ≥10%; P ≤ 0.05), with 502 genes showing H3K27ac losses and 448 genes showing H3K27ac gains (Supplementary Table 5).

GO enrichment analyses indicated that genes with significantly reduced H3K27ac in AD were enriched in cell junction organization and synaptic transmission (Figure 7A). Among genes with reduced CTCF binding (1241) and genes with reduced H3K27ac (502) in AD, a significant overlap (99 overlapped genes) was identified (Figure 7B, Supplementary Table 6), which represents 2.7-fold over-enrichment in overlapping genes. Moreover, the common genes are enriched in the regulation of synapse organization and membrane potential (Figure 7C, Supplementary Table 7). The top 30 overlapping hub genes were key players in synaptic organization and function, including NRXN1, HOMER1, CTNND2, and SV2B (Figure 7D). These data suggest that CTCF binding may affect the expression of synaptic genes via a mechanism involving the facilitation of histone acetylation, and the reduction of CTCF binding in AD leads to histone deacetylation and transcriptional reduction.

Figure 7, Genes with reduced H3K27ac overlap with genes with reduced CTCF binding in AD, and the common genes are enriched in synaptic organization.

Figure 7,

Open in a new tab

A, Gene Ontology (GO) showing the enriched pathways in Biological Processes (BP) of genes with decreased H3K27ac occupancy in AD patients (n=9), compared to control samples (n=9). B, Venn diagrams showing the overlap of genes with reduced CTCF binding and genes with decreased H3K27ac occupancy in AD. p = 3.18e-20, hypergeometric test. C, GO pathways of the overlapped genes with reduced CTCF binding and H3K27ac occupancy in AD. D, PPI of top 30 overlapped hub genes with reduced CTCF binding and H3K27ac occupancy in AD. Higher red intensities are associated with higher rankings.

4. Discussion

From proteomic studies of two longitudinal cohorts, it has been found that the increased abundance of synaptic genes is required for cognitive stability (Wingo et al., 2019). It suggests that the diminished expression of synaptic genes important for neuronal plasticity is responsible for cognitive decline in AD (Williams et al., 2021). For the discovery of therapeutic strategies to counteract AD progression, it is important to uncover how the transcription of synaptic genes is impaired in AD.

In chromosome of eukaryotes, DNA is wrapped around histones to form the nucleosomal fiber, which is further folded and looped into sophisticated higher-order three-dimensional structures. This 3D chromatin organization is controlled by CTCF, the master genome architecture protein that has crucial roles in the regulation of chromatin structure and gene transcription, including context-dependent promoter activation/repression, enhancer blocking and/or barrier insulation, and long-range chromatin interactions (Bell et al., 1999; de Wit et al., 2015; Dixon et al., 2012; Hou et al., 2008; Merkenschlager and Odom, 2013; Phillips and Corces, 2009). It has been found that CTCF binding at the proximal promoter facilitates distal enhancer-dependent gene activation (Kubo et al., 2021), thus, the alteration of CTCF occupancy at gene promoters could directly affect transcriptional outcomes.

Our epigenomic analyses of CTCF ChIP-seq data have identified a genome-wide reduction of CTCF binding in postmortem PFC of AD patients (n=9), compared to control subjects (n=9). Interestingly, genes with reduced CTCF binding in AD are enriched in synapse organization and cell junction assembly, including protocadherin (PCDH) and cadherin (CDH) family members, synaptic scaffolding/anchoring molecules or receptors such as SHANK2, GRIN2A, HOMER1, and NRXN1. These results have suggested a potential alteration of CTCF function in AD.

By comparing transcriptomic data from AD patients (Zhang et al., 2013), we have discovered that many of the adhesion and synaptic genes with reduced CTCF binding in AD are significantly diminished in their mRNA expression. For example, CTCF occupancies at the clustered protocadherin (PCDH) gene locus containing PCDHα, PCDHβ and PCDHγ were markedly reduced in AD patients, which was correlated with their transcriptional loss in AD. The expression of distinct repertoires of protocadherin protein isoforms serves as cell-surface molecular barcodes for neural circuit assembly (Canzio and Maniatis, 2019; Mountoufaris et al., 2018). CTCF is found to be heavily involved in promoter-enhancer interactions to regulate the expression of PCDH in neurons (Hirayama, 2012). Binding of the CTCF/cohesin complex and DNA looping to a distant enhancer is one of the mechanisms underlying stochastic expression of individual Pcdhα isoforms (Canzio and Maniatis, 2019), which generates enormous molecular diversity at the cell surface. The reduction of CTCF occupancy at PCDH gene locus in AD could lead to the disruption of molecular recognition codes and synaptic interactions among neurons.

Additional target genes with the significantly diminished CTCF binding and transcription in AD include GRIN2A and SHANK2, both of which are key components of the NMDAR complex involved in synaptic plasticity directly linked to cognition and memory. CTCF occupancy at the promoter and enhancer of GRIN2A and SHANK2 was significantly abrogated in AD, suggesting that these genes have lost CTCF-mediated chromatin looping to bring together promoters and enhancers, which could result in their diminished transcriptional activation in AD.

To better understand how the reduction of CTCF binding is linked to transcriptional changes in AD, we compared CTCF and H3K27ac ChIP-seq data in AD brains, as the increase of H3K27ac may play a role in the establishment of topologically associating domains (TADs) demarcated by CTCF during postmitotic transcriptional reactivation (Kang et al., 2020). Emerging evidence also implicates the alterations of histone acetylation and methylation in AD pathophysiology (Cao et al., 2020; Gräff et al., 2012; Nativio et al., 2020; Wang et al., 2021; Zheng et al., 2019). We have found altered H3K27ac peaks in AD patients, with both gains and losses (40.8% gains in peaks, 59.2% losses in peaks). It is slightly different from another epigenomic profiling, which found more gains than losses of H3K27ac peaks (57.6% gains, 42.4% losses) in AD (Nativio et al., 2020). The differences in human samples and selection criteria may underlie the discrepancy. We have identified a significant overlap of genes with the reduced CTCF binding and the reduced H3K27ac in AD, with the common genes enriched in synaptic organization. It is plausible that the reduced CTCF occupancy in AD changes the higher-order chromatin structure at specific genomic loci, causing the altered recruitment or binding of histone acetyltransferases or deacetylases, which results in the diminished transcriptional activation of target genes.

H3K4me3, an epigenetic mark associated with actively transcribing genes, plays a key role in regulating the expression of synaptic genes (Dincer et al., 2015). Thus, another epigenomic mechanism that may affect the transcription of synaptic genes in AD is the alteration of neuronal H3K4me3 epigenome. Genome-wide mapping has identified extensive H3K4me3 reorganization on neuronal genes at the early developmental stage in PFC (Cheung et al., 2010). Aberrant localization of H3K4me3 was found early in the course of AD (Mastroeni et al., 2015). Our discoveries on H3K4me3 alterations on genes regulating neuronal and synaptic functions in AD mouse models (Cao et al., 2020; Williams et al., 2023) also support the involvement of epigenetic abnormalities in neurodegenerative disorders.

Given the heterogeneity of human samples, the relatively small sample size of CTCF ChIP-seq data is a main limitation of this study, which only allows us to use a moderately stringent criterion in the differential CTCF peak analysis, as what has been used in a prior study for differential analyses of H3K27ac and H3K9ac peaks in control vs. AD patients (Nativio et al., 2020). With more epigenomic sequencing of a larger sample size, we will have more statistical power to determine what changes are more robust and what changes are not generally applicable.

In summary, this study has provided the first report revealing the binding changes of CTCF on genes involved in synaptic organization in AD human samples. Given the central role of CTCF in controlling genome architecture, it is possible that 3D chromatin organization at CTCF-binding genomic regions is altered in AD, which awaits to be tested in future studies using chromosome conformation capture techniques. Our finding has also provided a potential mechanism underlying the altered expression of synaptic genes in AD, which may directly contribute to synaptic dysfunction in this neurodegenerative disorder.

Supplementary Material

Sup Tables

Sup. Table 1: List of all peaks with significant alteration of CTCF binding in AD.

Sup. Table 2: GO enrichment analysis of genes with reduced CTCF binding in AD.

Sup. Table 3: SynGO analysis of synaptic genes with reduced CTCF binding in AD from top GO pathways.

Sup. Table 4: List of overlapping genes with reduced CTCF binding and DOWN expression in AD.

Sup. Table 5: List of genes with altered H3K27ac in AD.

Sup. Table 6: List of overlapping genes with CTCF loss and H3K27ac reduction in AD.

Sup. Table 7: GO enrichment analysis of overlapping genes with CTCF loss and H3K27ac reduction in AD.

Sup. Table 8: Information of human samples used in analyses of CTCF and H3K27ac ChIP-seq data.

Acknowledgements:

We are grateful to Dr. Jamal B. Williams for the introductory training of bioinformatic analysis packages. The work was supported by Center for Computational Research (CCR). We would also like to thank Drs. Ram Samudrala, Zackary Falls and their lab members for guiding us on how to use CCR resources and for statistical insights.

Fundings:

This work was supported by NIH grants AG064656 and AG079797 to Z.Y.

Footnotes

Disclosures:

The authors report no competing financial or other interests.

Data Availability:

The ChIP-seq data analyzed here have been deposited in the ENCODE public repository under accession codes listed in Supplementary Table 8, and can be accessed directly. The differential peak scores for CTCF and H3K27ac ChIP-seq data are deposited as matrix dataframes into a public repository.

References

  1. Barski A, et al. , 2007. High-resolution profiling of histone methylations in the human genome. Cell. 129, 823–37. [DOI] [PubMed] [Google Scholar]
  2. Bell AC, et al. , 1999. The protein CTCF is required for the enhancer blocking activity of vertebrate insulators. Cell. 98, 387–96. [DOI] [PubMed] [Google Scholar]
  3. Borrelli E, et al. , 2008. Decoding the epigenetic language of neuronal plasticity. Neuron. 60, 961–74. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Canzio D, Maniatis T, 2019. The generation of a protocadherin cell-surface recognition code for neural circuit assembly. Curr Opin Neurobiol. 59, 213–220. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Cao Q, et al. , 2020. Targeting histone K4 trimethylation for treatment of cognitive and synaptic deficits in mouse models of Alzheimer’s disease. Sci Adv. 6. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Chen Y, et al. , 2016. From reads to genes to pathways: differential expression analysis of RNA-Seq experiments using Rsubread and the edgeR quasi-likelihood pipeline. F1000Research. 5, 1438. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Chernukhin I, et al. , 2007. CTCF interacts with and recruits the largest subunit of RNA polymerase II to CTCF target sites genome-wide. Mol Cell Biol. 27, 1631–48. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Cheung I, et al. , 2010. Developmental regulation and individual differences of neuronal H3K4me3 epigenomes in the prefrontal cortex. Proc Natl Acad Sci U S A. 107, 8824–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Chin C-H, et al. , 2014. cytoHubba: identifying hub objects and sub-networks from complex interactome. BMC Systems Biology. 8, S11. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Creyghton MP, et al. , 2010. Histone H3K27ac separates active from poised enhancers and predicts developmental state. Proceedings of the National Academy of Sciences. 107, 21931–21936. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. De Leeuw CA, et al. , 2015. MAGMA: Generalized Gene-Set Analysis of GWAS Data. PLOS Computational Biology. 11, e1004219. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. de Wit E, et al. , 2015. CTCF Binding Polarity Determines Chromatin Looping. Mol Cell. 60, 676–84. [DOI] [PubMed] [Google Scholar]
  13. Dehingia B, et al. , 2022. CTCF shapes chromatin structure and gene expression in health and disease. EMBO Rep. 23, e55146. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Dincer A, et al. , 2015. Deciphering H3K4me3 broad domains associated with gene-regulatory networks and conserved epigenomic landscapes in the human brain. Transl Psychiatry. 5, e679. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Dixon JR, et al. , 2012. Topological domains in mammalian genomes identified by analysis of chromatin interactions. Nature. 485, 376–80. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. ENCODE Project Consortium, 2012. An integrated encyclopedia of DNA elements in the human genome. Nature. 489, 57–74. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Engel N, et al. , 2006. CTCF binding sites promote transcription initiation and prevent DNA methylation on the maternal allele at the imprinted H19/Igf2 locus. Hum Mol Genet. 15, 2945–54. [DOI] [PubMed] [Google Scholar]
  18. Finucane HK, et al. , 2015. Partitioning heritability by functional annotation using genome-wide association summary statistics. Nature Genetics. 47, 1228–1235. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Fraser P, Bickmore W, 2007. Nuclear organization of the genome and the potential for gene regulation. Nature. 447, 413–7. [DOI] [PubMed] [Google Scholar]
  20. Gazal S, et al. , 2017. Linkage disequilibrium–dependent architecture of human complex traits shows action of negative selection. Nature Genetics. 49, 1421–1427. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Gräff J, et al. , 2012. An epigenetic blockade of cognitive functions in the neurodegenerating brain. Nature. 483, 222–6. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Heberle H, et al. , 2015. InteractiVenn: a web-based tool for the analysis of sets through Venn diagrams. BMC Bioinformatics. 16. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Hirayama T, 2012. CTCF Is Required for Neural Development and Stochastic Expression of Clustered Pcdh Genes in Neurons. Cell. [DOI] [PubMed] [Google Scholar]
  24. Hou C, et al. , 2008. CTCF-dependent enhancer-blocking by alternative chromatin loop formation. Proc Natl Acad Sci U S A. 105, 20398–403. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Kang H, et al. , 2020. Dynamic regulation of histone modifications and long-range chromosomal interactions during postmitotic transcriptional reactivation. Genes Dev. 34, 913–930. [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Konrad EDH, et al. , 2019. CTCF variants in 39individuals with a variable neurodevelopmental disorder broaden the mutational andclinical spectrum. Genetics in Medicine. 21, 2723–2733. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Koopmans F, et al. , 2019. SynGO: An Evidence-Based, Expert-Curated Knowledge Base for the Synapse. Neuron. 103, 217–234.e4. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Kubo N, et al. , 2021. Promoter-proximal CTCF binding promotes distal enhancer-dependent gene activation. Nature Structural & Molecular Biology. 28, 152–161. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Lanctôt C, et al. , 2007. Dynamic genome architecture in the nuclear space: regulation of gene expression in three dimensions. Nat Rev Genet. 8, 104–15. [DOI] [PubMed] [Google Scholar]
  30. Lieberman-Aiden E, et al. , 2009. Comprehensive mapping of long-range interactions reveals folding principles of the human genome. Science. 326, 289–93. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. López-Otín C, et al. , 2023. Hallmarks of aging: An expanding universe. Cell. 186, 243–278. [DOI] [PubMed] [Google Scholar]
  32. Luo Y, et al. , 2020. New developments on the Encyclopedia of DNA Elements (ENCODE) data portal. Nucleic Acids Res. 48, D882–D889. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Mastroeni D, et al. , 2015. Aberrant intracellular localization of H3k4me3 demonstrates an early epigenetic phenomenon in Alzheimer’s disease. Neurobiol Aging. 36, 3121–3129. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Mathys H, et al. , 2019. Single-cell transcriptomic analysis of Alzheimer’s disease. Nature. 570, 332–337. [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. McCarthy DJ, et al. , 2012. Differential expression analysis of multifactor RNA-Seq experiments with respect to biological variation. Nucleic Acids Res. 40, 4288–97. [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Merkenschlager M, Duncan, 2013. CTCF and Cohesin: Linking Gene Regulatory Elements with Their Targets. Cell. 152, 1285–1297. [DOI] [PubMed] [Google Scholar]
  37. Merkenschlager M, Odom DT, 2013. CTCF and cohesin: linking gene regulatory elements with their targets. Cell. 152, 1285–97. [DOI] [PubMed] [Google Scholar]
  38. Morabito S, et al. , 2020. Integrative genomics approach identifies conserved transcriptomic networks in Alzheimer’s disease. Human Molecular Genetics. 29, 2899–2919. [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Mountoufaris G, et al. , 2018. Writing, Reading, and Translating the Clustered Protocadherin Cell Surface Recognition Code for Neural Circuit Assembly. Annu Rev Cell Dev Biol. 34, 471–493. [DOI] [PubMed] [Google Scholar]
  40. Nativio R, et al. , 2020. An integrated multi-omics approach identifies epigenetic alterations associated with Alzheimer’s disease. Nat Genet. 52, 1024–1035. [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Nora EP, et al. , 2017. Targeted Degradation of CTCF Decouples Local Insulation of Chromosome Domains from Genomic Compartmentalization. Cell. 169, 930–944.e22. [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. Ohlsson R, et al. , 2010. Does CTCF mediate between nuclear organization and gene expression? BioEssays. 32, 37–50. [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Ohlsson R, et al. , 2001. CTCF is a uniquely versatile transcription regulator linked to epigenetics and disease. Trends Genet. 17, 520–7. [DOI] [PubMed] [Google Scholar]
  44. Ong C-T, Corces VG, 2014. CTCF: an architectural protein bridging genome topology and function. Nature Reviews Genetics. 15, 234–246. [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. Phillips JE, Corces VG, 2009. CTCF: master weaver of the genome. Cell. 137, 1194–211. [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. Quinlan AR, Hall IM, 2010. BEDTools: a flexible suite of utilities for comparing genomic features. Bioinformatics. 26, 841–842. [DOI] [PMC free article] [PubMed] [Google Scholar]
  47. Ramírez F, et al. , 2014. deepTools: a flexible platform for exploring deep-sequencing data. Nucleic Acids Res. 42, W187–91. [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. Rao SS, et al. , 2014. A 3D map of the human genome at kilobase resolution reveals principles of chromatin looping. Cell. 159, 1665–80. [DOI] [PMC free article] [PubMed] [Google Scholar]
  49. Robinson JT, et al. , 2011. Integrative genomics viewer. Nature Biotechnology. 29, 24–26. [DOI] [PMC free article] [PubMed] [Google Scholar]
  50. Robinson MD, et al. , 2010. edgeR: a Bioconductor package for differential expression analysis of digital gene expression data. Bioinformatics. 26, 139–40. [DOI] [PMC free article] [PubMed] [Google Scholar]
  51. Ross-Innes CS, et al. , 2012. Differential oestrogen receptor binding is associated with clinical outcome in breast cancer. Nature. 481, 389–393. [DOI] [PMC free article] [PubMed] [Google Scholar]
  52. Selkoe DJ, 2002. Alzheimer’s disease is a synaptic failure. Science. 298, 789–91. [DOI] [PubMed] [Google Scholar]
  53. Shannon P, et al. , 2003. Cytoscape: A Software Environment for Integrated Models of Biomolecular Interaction Networks. Genome Research. 13, 2498–2504. [DOI] [PMC free article] [PubMed] [Google Scholar]
  54. Sharifi-Zarchi A, et al. , 2017. DNA methylation regulates discrimination of enhancers from promoters through a H3K4me1-H3K4me3 seesaw mechanism. BMC Genomics. 18. [DOI] [PMC free article] [PubMed] [Google Scholar]
  55. Szklarczyk D, et al. , 2021. The STRING database in 2021: customizable protein–protein networks, and functional characterization of user-uploaded gene/measurement sets. Nucleic Acids Research. 49, D605–D612. [DOI] [PMC free article] [PubMed] [Google Scholar]
  56. Tracy TE, Gan L, 2018. Tau-mediated synaptic and neuronal dysfunction in neurodegenerative disease. Curr Opin Neurobiol. 51, 134–138. [DOI] [PMC free article] [PubMed] [Google Scholar]
  57. Vostrov AA, Quitschke WW, 1997. The zinc finger protein CTCF binds to the APBbeta domain of the amyloid beta-protein precursor promoter. Evidence for a role in transcriptional activation. J Biol Chem. 272, 33353–9. [DOI] [PubMed] [Google Scholar]
  58. Wang Q, et al. , 2022. Exploring Epigenomic Datasets by ChIPseeker. Current Protocols. 2, e585. [DOI] [PubMed] [Google Scholar]
  59. Wang W, et al. , 2021. Epigenetic treatment of behavioral and physiological deficits in a tauopathy mouse model. Aging Cell. 20, e13456. [DOI] [PMC free article] [PubMed] [Google Scholar]
  60. Wang Y, et al. , 2018. The 3D Genome Browser: a web-based browser for visualizing 3D genome organization and long-range chromatin interactions. Genome Biology. 19. [DOI] [PMC free article] [PubMed] [Google Scholar]
  61. Williams JB, et al. , 2023. Inhibition of histone methyltransferase Smyd3 rescues NMDAR and cognitive deficits in a tauopathy mouse model. Nat Commun. 14, 91. [DOI] [PMC free article] [PubMed] [Google Scholar]
  62. Williams JB, et al. , 2021. Transcriptomic analysis of human brains with Alzheimer’s disease reveals the altered expression of synaptic genes linked to cognitive deficits. Brain Commun. 3, fcab123. [DOI] [PMC free article] [PubMed] [Google Scholar]
  63. Wilson DM 3rd, et al. , 2023. Hallmarks of neurodegenerative diseases. Cell. 186, 693–714. [DOI] [PubMed] [Google Scholar]
  64. Wingo AP, et al. , 2019. Large-scale proteomic analysis of human brain identifies proteins associated with cognitive trajectory in advanced age. Nature Communications. 10, 1619. [DOI] [PMC free article] [PubMed] [Google Scholar]
  65. Xie Z, et al. , 2007. Kalirin-7 controls activity-dependent structural and functional plasticity of dendritic spines. Neuron. 56, 640–56. [DOI] [PMC free article] [PubMed] [Google Scholar]
  66. Yang JH, et al. , 2023. Loss of epigenetic information as a cause of mammalian aging. Cell. 186, 305–326.e27. [DOI] [PMC free article] [PubMed] [Google Scholar]
  67. Yu G, et al. , 2015. ChIPseeker: an R/Bioconductor package for ChIP peak annotation, comparison and visualization. Bioinformatics. 31, 2382–3. [DOI] [PubMed] [Google Scholar]
  68. Zhang B, et al. , 2013. Integrated systems approach identifies genetic nodes and networks in late-onset Alzheimer’s disease. Cell. 153, 707–20. [DOI] [PMC free article] [PubMed] [Google Scholar]
  69. Zheng Y, et al. , 2019. Inhibition of EHMT1/2 rescues synaptic and cognitive functions for Alzheimer’s disease. Brain. 142, 787–807. [DOI] [PMC free article] [PubMed] [Google Scholar]
  70. Zhou Y, et al. , 2019. Metascape provides a biologist-oriented resource for the analysis of systems-level datasets. Nature Communications. 10. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Sup Tables

Sup. Table 1: List of all peaks with significant alteration of CTCF binding in AD.

Sup. Table 2: GO enrichment analysis of genes with reduced CTCF binding in AD.

Sup. Table 3: SynGO analysis of synaptic genes with reduced CTCF binding in AD from top GO pathways.

Sup. Table 4: List of overlapping genes with reduced CTCF binding and DOWN expression in AD.

Sup. Table 5: List of genes with altered H3K27ac in AD.

Sup. Table 6: List of overlapping genes with CTCF loss and H3K27ac reduction in AD.

Sup. Table 7: GO enrichment analysis of overlapping genes with CTCF loss and H3K27ac reduction in AD.

Sup. Table 8: Information of human samples used in analyses of CTCF and H3K27ac ChIP-seq data.

NIHMS1931430-supplement-Sup_Tables.xlsx (568KB, xlsx)

Data Availability Statement

The ChIP-seq data analyzed here have been deposited in the ENCODE public repository under accession codes listed in Supplementary Table 8, and can be accessed directly. The differential peak scores for CTCF and H3K27ac ChIP-seq data are deposited as matrix dataframes into a public repository.

RESOURCES